INTERNATIONAL JOURNAL ON SMART SENSING AND INTELLIGENT SYSTEMS VOL. 8, NO. 1, MARCH 2015 90 A NOVEL APPROACH OF COMBINING STEGANOGRAPHY ALGORITHMS Hamdan Lateef Jaheel and Zou Beiji School of Information Science and Engineering, Central South University, Hunan, P.R. China Submitted: Nov. 3, 2014 Accepted: Jan. 5, 2015 Published: Mar. 1, 2015 Abstract- Steganography is the act of hiding a message inside another message in such a way that the hidden message can only be detected by its intended recipient. In this paper, we combined two steganography algorithms namely JSteg and OutGuess algorithms, in order to exploit the beneficial characteristics and features of both algorithms to enhance the protection level for secret images. In our proposed approach, the secret message (image) is first concealed inside another image using JSteg algorithm and the resultant stego-image is further hidden inside a final image using OutGuess 0.1 algorithm. In this combine approach, the tricky nature of hiding an already hidden message is using two different algorithms increases the level of difficulty for a third party to suspect the existence of a secret image in the first place or even successful decode the it. Besides that, the priority given to the choice of a good image size and type in this approach further disguises the secret image and increases the chances that the image could go unnoticed. Results after calculating the capacity and PSNR for images proved that our approach is a good and acceptable steganography system. The model presented here is based on JPEG images. Index terms: Transform Domain Technique, Jsteg, OutGuess0.1, MSE , peak-signal-to-noise ratio (PSNR).
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
INTERNATIONAL JOURNAL ON SMART SENSING AND INTELLIGENT SYSTEMS VOL. 8, NO. 1, MARCH 2015
90
A NOVEL APPROACH OF COMBINING STEGANOGRAPHY
ALGORITHMS
Hamdan Lateef Jaheel and Zou Beiji
School of Information Science and Engineering, Central South University,
Hamdan L. Jaheel and Zou Beiji, A NOVEL APPROACH OF COMBINING STEGANOGRAPHY ALGORITHMS
97
Step1: read cover image1. JPEG
A. JPEG partitions a cover image1 into non overlapping blocks of 8*8 pixels
B. Calculate DCT coefficient for each block
C. Quantize the coefficients
Step2: hiding process by using JSteg algorithm While left to embed do
A. Get next DCT coefficients from cover image1
B. If DCT ≠0, DCT ≠1 & DCT ≠ -1 then
C. Get LSB from the message
D. Replace DCT LSB with message bit
End (if)
End (while)
Step3: calculate message capacity
Step 4: Writ JPEG image by de-quantize and take inverse DCT to obtain stego image1
Secret image2= stego image1
Step5: Read cover image2.JPEG
A. JPEG partitions a cover image2 into non overlapping blocks of 8*8 pixels
B. Calculate DCT coefficient for each block
C. Quantize the coefficients
Step6:hiding process by using Outguess algorithm While left to embed do
A. Get pseudo random DCT coefficient from cover image2
B. If DCT ≠0, DCT ≠1 & DCT ≠ -1 then
C. Get LSB from the message
D. Replace DCT LSB with message bit
End (if)
End (while)
Step7: calculate message capacity
Step8: Write JPEG image by de-quantize and take inverse DCT to obtain stego image2.
The algorithm was implementation on Matlab 7.6 platform the results are shown in Figure
(2) .And from the result we can see that the proposed approach successfully combined two
steganographic methods in frequency domain, where an intended secret image (hidden image1) is first hidden using JSteg algorithm and the resultant image is again hidden in another image using
OutGuess algorithm.
INTERNATIONAL JOURNAL ON SMART SENSING AND INTELLIGENT SYSTEMS VOL. 8, NO. 1, MARCH 2015
98
. Fig2: Explain encoding process
6.2 Extracting algorithm
Input: Stego image2
Step1: read Stego image2.JPEG
A. JPEG partitions Stego image2 into non overlapping blocks of 8*8 pixels
B. Calculate DCT coefficient for each block C. Quantize the coefficients D. Calculate message capacity
Step3: Extracting process by using Outguess algorithm
While left to embed do
A. Get pseudo random DCT coefficient from Stego image2
B. If DCT ≠0, DCT ≠1 & DCT ≠ -1 then
C. Get LSB from the message
D. Replace DCT LSB with message bit End (if) End (while)
Step4: Writ JPEG image by de-quantize and take inverse DCT to obtain secret image2.
Stego image1= Secret image2
Step5: Read Stego image1.JPEG
A. JPEG partitions Stego image1 into non overlapping blocks of 8*8 pixels
B. Calculate DCT coefficient for each block C. Quantize the coefficients D. Calculate message capacity
Step6: Extracting process by using JSteg algorithm
While left to embed do
A. Get next DCT coefficients from Stego image1
B. If DCT ≠0, DCT ≠1 & DCT ≠ -1 then
+
+
Hamdan L. Jaheel and Zou Beiji, A NOVEL APPROACH OF COMBINING STEGANOGRAPHY ALGORITHMS
99
C. Concatenate DCT LSB to secret message
End (if) End (while)
Step7: Write JPEG image by de-quantize and take inverse DCT to obtain secret image1.
After the implementation of this algorithm in Matlab 7.6 program the results obtained are shown
in Figure (3):
Fig3: Explain decoding process
VII. EXPERIMENTAL AND RESULTS
The experiments were implemented on a set of images downloaded from the images
database at Washington university [23, 28, 29] and Oklahoma University [24, 30] (more than 500
images of JPEG type) and also some images from the special camera. Fundamental information
hiding systems: capacity, security, and durability. The capacity is the amount of data that is
possible to be hidden in a cover medium. Security refers to the inability of the attacker to detect
hidden data. Robustness refers to the extent to which the stego medium can withstand the attacker,
which can destroy the hidden information.
Embedding Capacity
It is the maximum size of the secret data that can be embedded in the cover image without
deteriorating the integrity of the cover image. It can be represented in bytes or Bit Per
Pixel(bpp),The calculated explain in equation 1.
capacity =(X*Y)/64 * b *(n − 15) (1)
In this equation, X and Y are the dimensions of the cover image. By dividing the product
of X, Y by 64, the number of 8*8 blocks is achieved. During data embedding process, no data are
INTERNATIONAL JOURNAL ON SMART SENSING AND INTELLIGENT SYSTEMS VOL. 8, NO. 1, MARCH 2015
100
embedded in the last 15 coefficients, so the term (n-15) is used here, and in each coefficient b bits
of data will be embedded.
Peak-signal-to-noise ratio (PSNR)
As a performance measurement for image distortion, the well known peak-signal-to-noise
ratio (PSNR) which is classified under the difference distortion metrics is applied to the stego-
images. It is defined as Eq (2):
���� = 10 log 10 �����
�
����(2)
Where MSE denotes mean square error which is given as Eq (3):
�� =1
������ − ��� �(3)�
�
�
�
Table (1) illustrates the capacity and PSNR for the encoding process (1).
Table1.Capacity and PSNR for images
Image Capacity PSNR
Encoding
process1
Outguess stego-image
Stego 1043768 56.8607
Where x and y are the image coordinates, M and N are the dimensions of the image, Sxv is the
generated stego-image and Cxv is the cover image. Also����� holds the maximum value in the
image, for example:
����� ≤ � 1,
255,
�������������������8���
Many authors [19 - 27],consider Cmax=255 as a default value for 8 bit images. It can be
the case, for instance, that the examined image has only up to 253 or fewer presentations of gray
colors. Knowing that C���� results in a severe change to the PSNR value. This Cmax can be
defined as the actual maximum value rather than the largest possible value. PSNR is often
expressed on a logarithmic scale in decbels (dB). PSNR values falling below 30 dB indicate a
fairly low quality, i.e., distortion caused by embedding can be obvious; however, a high quality
stego-image should strive for 40 dB and above. In this paper, after combining two concealment
algorithms specifically JSteg and OutGuess algorithm. We use the OutGuess 0.1 algorithm to
further enhance the level of protection for detection of a secret message (image), which has
already been hidden inside another image using an Jsteg algorithm. And as such, the tricky nature
Hamdan L. Jaheel and Zou Beiji, A NOVEL APPROACH OF COMBINING STEGANOGRAPHY ALGORITHMS
101
of hiding an already hidden image is using two different algorithms introduces some complexity
and makes it more deceptive to a third parties.This in effect reduces the suspension in the
existence of the secret image, thereby significantly enhancing the image protection level .
The hidden image (size image <130×130) will be stored two times. The first time by using
jsteg algorithm and the second time hiding the resultant image (Jsteg stego) using outguess
algorithm. The capacity is calculated two times to get the hidden image. The first time for Jsteg
and the second time for OutGuess and this is another factor adding safety to the secret image.
The results of the PSNR values of the Stego-images calculated after sending the final Stego-
images via e-mail to another computer, and retrieving the hidden messages (image), were
between (50-57)db and this range is considered to be a very good and acceptable steganography
system. Fig(5) is an illustration of the encoding processes, and as shown in table(2) explaining
the PSNR & capacity for some encoding processes. Fig(5) explain coefficient histogram for
cover image and final stego image.
Secret image1 Stego image1 stego image2
Fig4: illustrate for some of encoding processes (a) encoding process1, (b) encoding process2,
INTERNATIONAL JOURNAL ON SMART SENSING AND INTELLIGENT SYSTEMS VOL. 8, NO. 1, MARCH 2015
102
Fig(6): Coefficient histogram for (A) cover image (B) outguess
steganograme
(c)
encoding process
3and (d) encoding process 4.
Table2. Explain PSNR & capacity for some encoding processes.
Encoding
process Image Capacity PSNR
A
Outguess
stego-
image
1034728 57.8352
B Outguess stego-
image
1325120 59.6505
Hamdan L. Jaheel and Zou Beiji, A NOVEL APPROACH OF COMBINING STEGANOGRAPHY ALGORITHMS
103
C
Outguess
stego-
image
1037432 55.8158
D
Outguess
stego-
image
2546416 54.3120
Table (3) show that PSNR of our proposed technique is better than proposed technique in
reference [25-27], because the PSNR value of our technique exceed that of the previous
technique with a significant margin. Depending on the result of the comparison, we find that the
proposed method is good and acceptable and safe steganography scheme.
Table(3):comparison between our proposed method with the results of technique in refrencep[25].
Cover image Previous results
proposed
method
PSNR (dB) PSNR (dB)
Lena 41.79 53.1257
Baboon 37.90 50.0200
Airplane 40.60 53.0196
Peppers 40.97 53.1825
VIII. CONCLUSIONS
This paper presented a steganographic approach that combined jsteg and outguess
algorithms. The approach allowed us to benefit from the potential features and strengths of both
algorithms and this added a significant level of protection to hidden images. In principle what
happened in our proposed approach is that an image intended to be a secret image is first hidden
in an image using jsteg algorithm and the resultant stego image is further hidden in another
INTERNATIONAL JOURNAL ON SMART SENSING AND INTELLIGENT SYSTEMS VOL. 8, NO. 1, MARCH 2015
104
second image using outguess 0.1 algorithm to produce a final stego image. The act of hiding an
already hidden image (stego image) in another image alone is tricky and deceptive for a third
party. Besides that, the idea of combining two steganographic algorithm makes the approach
more complex for a third party and this increases the chances that the intended secret massage
(secret image) could go unnoticed.
Furthermore, the priority given to selecting a good image sizes and type further disguises
the secret image and makes it more difficult for a third party to suspect the existence of a secret
image. The experimental results indicated an average PSNR value of more than 50 dB for more
than100 images and that is a good and acceptable steganography scheme. As future work, we
could try the combination of other steganography techniques and compare the efficiency levels,
as well as adding image encryption.
REFERENCES
[1] Niels Provos and Peter Honeyman “Hide and Seek: An Introduction to Steganography” ,
IEEE Computer Society ,Vol.1,No.3, 2003,pp.32-44.
[2] Philip Bateman and Dr. Hans “Image Steganography and Steganalysis”, M.S., Department of
Computing Faculty of Engineering and Physical Sciences, University of Surrey Guildford Surrey,